5/19/2022
Mox/Hyak is up and running!
Setup for categorical covariates (i.e., natal origin)
For our simulation, with three natal origins (JDR, YAK, and TUC)
Parameters:
- Grand mean = \(\beta_0\)
- Origin effect, JDR relative to TUC = \(\beta_{origin,1}\)
- Origin effect, YAK relative to TUC = \(\beta_{origin,2}\)
Effect of origin:
\[ JDR: \beta_0 + \beta_{origin,1} \\ YAK: \beta_0 + \beta_{origin,2} \\ TUC: \beta_0 - \beta_{origin,1} - \beta_{origin,2} \]
Interpretation is a bit funky
For example, in order for JDR origin fish to have a different movement probability than YAK and TUC origin fish, the following relationship must be followed:
\[ YAK = TUC \\ \beta_0 + \beta_{origin,2} = \beta_0 - \beta_{origin,1} - \beta_{origin,2} \\ 2 * \beta_{origin,2} = - \beta_{origin,1} \]
For example, \(\beta_0\) = 1, \(\beta_{origin,1}\) = -0.5 and \(\beta_{origin,2}\) = 0.25 would lead to these observed effects:
\[ JDR = 0.5 \\ YAK = 1.25 \\ TUC = 1.25 \]
Simulation model results
Origin only: b0
Origin only: borigin1
Origin only: borigin2
Run 4 - simulated dataset
## origin rear state2 n indices ## 1 1 1 loss 5 [8,10] ## 2 1 2 loss 5 [8,10] ## 3 2 1 loss 3 [8,10] ## 4 2 2 loss 1 [8,10] ## 5 3 1 loss 34 [8,10] ## 6 3 2 loss 43 [8,10] ## 7 1 1 mainstem, ICH to LGR 15 [8,5] ## 8 1 2 mainstem, ICH to LGR 20 [8,5] ## 9 3 1 mainstem, ICH to LGR 45 [8,5] ## 10 3 2 mainstem, ICH to LGR 74 [8,5]
## origin rear state2 n indices ## 1 1 1 loss 3 [8,10] ## 2 1 2 loss 3 [8,10] ## 3 2 1 loss 2 [8,10] ## 4 2 2 loss 2 [8,10] ## 5 3 1 loss 24 [8,10] ## 6 3 2 loss 40 [8,10] ## 7 1 1 mainstem, ICH to LGR 16 [8,5] ## 8 1 2 mainstem, ICH to LGR 24 [8,5] ## 9 2 1 mainstem, ICH to LGR 3 [8,5] ## 10 2 2 mainstem, ICH to LGR 4 [8,5] ## 11 3 1 mainstem, ICH to LGR 46 [8,5] ## 12 3 2 mainstem, ICH to LGR 41 [8,5]
Origin + rear: beta_0
Origin + rear: rear
Origin + rear: beta_origin1
Origin + rear: beta_origin2
Temperature only: beta_0
Temperature only: beta_temp
Temperature + flow - correlation issues?
Temperature + flow: b0
Temperature + flow: btemp
Temperature + flow: bflow
Temperature, flow, origin, and rear
Spill covariate
- Binary variable (was there any spill around a window of time of the estimated transition date), or count of days of spill within window?